Moving disks from DL360 G2 to DL360 G3 - mirroring doesn't work

[Sigurd Erland]

I have a DL360 G2 server, set up with a mirror set of two 18GB disks. I shutdown the server, take out the first disk (disk0), and insert it as the only disk in a DL360 G3 server, and then I boot the G3 server. I get a message saying that one disk is missing, and the server is running in Interim Recovery mode, just as expected. Both servers are using the 5i controller, so there is no problem starting Windows. So far, everything looks fine.

BUT: When I insert a second disk in the G3, to rebuild the mirror set, the synchronization doesn't start as it should. Normally, the new disk should start getting sync'ed with the existing disk after a few seconds. Instead, the new disk lights up a bit, and after a few seconds, it goes black, with no LED's lit. I have tried with several different disks, both the same size as the original, and bigger ones, both Ultra 3 disks and U320 disks. No difference.

I checked the firmware on the 5i controller in the G3 server, it has the latest firmware.

Is there some general problem moving disks between servers that could break the mirror sync. mechanism?

[James ~ Happy Dude]

Hello Sigurd,
What do you see in the Array Configuration Utility ?
If not.. boot from the smartstart CD check the status of the RAID.

Regards.

[Sigurd Erland]

In the ACU, in "logical view", there is first a "36,4 GB Parallell SCSI Unassigned Drive at Port 1: SCSI ID 1". Below that, there is an Array A, marked with a red cross, under this array there is a "Logical Drive 1 (17359 MB, RAID 1+0), and also a "Unused Space, ???".

If I go to "physical view", there is first a "36,4 GB Parallell SCSI Unassigned Drive at Port 1: SCSI ID 1". Below that, there is an Array A, marked with a red cross, under this array there is a "18,2 GB Parallell SCSI Drive at Port 1: SCSI ID 0", and also a "??? Parallell SCSI Drive at Port 2: SCSI ID 1". The last one is marked with a red cross.

[KarloChacon]

hi Sigurd

if you can post a screenshot from the ACU and also a report from Array Diagnostic Utility will be helpful to check what is going on


regards

[Sigurd Erland]

Here is a zip file containing a screenshot from the ACU, and also the output from the ADU.

[dms_1]

Hi Sigurd,

from the ADU report:

ErrorText SLOT 0 Smart Array 5i Controller
ErrorText Logical drive 1 status = Interim recovery (volume functional, but not fault tolerant)
ErrorText SCSI Port 2, drive ID 1 failed - REPLACE
ErrorText SCSI Port 1 Drive ID 1 RIS copies within this drive do not match
ErrorText Error occurred reading RIS copy from SCSI Port 2 Drive ID 1


So the RIS tabbles of the HDDs are corrupted.RIS tables on the HDDs contain information about the RAID settings.Old HDDs may contain RIS table, which may confuse the controller.Your controller is searching for a HDD on SCSI Port 2, drive ID 1.
You can try booting the server once without HDDs at all - that will reset the controller.After that boot the server with both HDD and check if rebuild starts.

I noticed that the new HDD (SCSI Port 1, Drive ID 1 ) has an old Firmware Revision (HPB1). If possible update HDD firmware from the FW CD or from the this link:

http://h18007.www1.hp.com/support/files/storage/us/index.html


The best way to overwrite RIS tables should be to backup the data, delete the array, recreate it and restore from backup after that...

[Sigurd Erland]

Thanks for the suggestion. I will try this first thing tomorrow.

[Sigurd Erland]

I have now tried deleting the array on the server, but no progress.

However, after looking a bit more on this, I believe that the problem is probably that the 18GB disk I have taken from a DL380G2 has been connected to SCSI Port 2 in the DL380G2 (which is the only port we can use there).

Sorry, I see that in my first post, I said that the original server was a DL360G2. My mistake.

When I put the disk into the DL360G3, the disks are connected to SCSI Port 1, and when the RAID info on the disk is read, the server expects to find the missing drive on SCSI Port 2, ID1, which it of course cannot.

I could probably solve the problem as such by erasing the drive, thus resetting the RAID info on it, but since the whole idea is that I wants to keep the data on the disk, this wouldn't solve anything.

Is there any way of editing / changing the RAID info on the disk without erasing the data?

Or, is there any other way I can solve this?

[Brian_Murdoch]

Hi Sigurd,

This is quite common when moving a single drive from a system where the default port is port 2. Unfortunately there is not any quick solution. I believe that one way round it is as follows:-

On DL380 G2 insert a new fresh disk in the disk 0 position which was removed and let it rebuild completely. Shut down and now remove the original member of the mirrored pair (disk 1). Make sure the system will now reboot OK with the replacement disk before doing anything else. Once it has booted OK place the removed disk back with it's original partner on the DL360 G3.

It's not only the port information which is held on RIS, the drive serial numbers and other information is there too. Hopefully when the controller detects that both disks are the original members of the mirror, it should not complain since "drive roaming" is a feature of the Smart Array controllers.
In other words you can move disks from internal bays to external bays and to other servers providing they remain intact.

I hope this works.

Regards,

Brian

[Sigurd Erland]

Thanks Brian, but in the meantime I got tired of waiting for a quick fix, and went for the manual reinstall of the new server (OS, hotfixes, and Citrix). I'll keep this in mind, however, if I need to do something like this again.